Retinal Electrophysiological Patterns in Alzheimer's Disease: A Multi-Domain Signal Processing Framework for Non-Invasive Biomarker Discovery Using a Portable ERG Device

Abstract

Alzheimer’s disease (AD) is a widespread neurodegenerative disorder that currently lacks early and accessible diagnostic tools. While the electroretinogram (ERG) provides a non-invasive way to detect retinal dysfunction linked to neurodegeneration, it has been unclear whether reliable biomarkers can be derived beyond traditional amplitude and latency measurements. Here, we implemented a multi-domain signal processing framework to analyze ERG signals from 46 participants (20 with AD and 26 controls) using a portable, handheld device (RETeval). The framework consists of five complementary techniques: multiscale fuzzy entropy, FFT harmonic analysis, stimulus-response wavelet time-frequency coherence, a novel inter-cycle lag variant of sample entropy, and discrete wavelet transform. We identified seven significant candidate biomarkers, five of which showed large effect sizes. A logistic regression classifier combining three of these biomarkers achieved an ROC-AUC of 0.858, with 70.0% sensitivity and 88.5% specificity. These findings suggest that multi-domain ERG analysis captures retinal temporal dysfunction signatures in AD patients that are not accessible via standard clinical analysis, supporting the use of portable ERG devices as a promising, non-invasive tool for early AD detection.

Joel Barría

Postdoctoral Researcher

Leo Medina

Principal Investigator

Leo teaches engineering courses at Usach, and his research interests are in the neural engineering and computational neuroscience fields. His work has contributed to understand how nerve fibers respond to electrical stimulation.